See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/5499756

Quantitative determination of some food dyes

using digital processing of images obtained by
thin-layer chromatography

ARTICLE in JOURNAL OF CHROMATOGRAPHY A MAY 2008

Impact Factor: 4.26 DOI: 10.1016/j.chroma.2008.02.077 Source: PubMed

CITATIONS

33

3 AUTHORS, INCLUDING:

Augustin C. Mot Costel Srbu

Babe-Bolyai University Babe-Bolyai University
35 PUBLICATIONS 213 CITATIONS 131 PUBLICATIONS 964 CITATIONS

SEE PROFILE SEE PROFILE

Available from: Costel Srbu

Retrieved on: 27 August 2015
 Available online at www.sciencedirect.com

Journal of Chromatography A, 1188 (2008) 295300

Quantitative determination of some food dyes using digital processing

of images obtained by thin-layer chromatography
Florin Soponar, Augustin Catalin Mot, Costel Sarbu
Babes-Bolyai University, Faculty of Chemistry and Chemical Engineering,
Arany Janos Street 11, RO-400028 Cluj Napoca, Romania
Received 13 December 2007; received in revised form 10 February 2008; accepted 12 February 2008
Available online 29 February 2008

Abstract
A high-performance thin-layer chromatographic method combined with image processing of scanned chromatograms was developed for the
determination of some food dyes (tartrazine, azorubine and Sunset Yellow) in different products. Porous silica gel with 3-aminopropyl functional
groups attached to the matrix was used as stationary phase and a mixture of isopropanol, diethyl ether and ammonia (2:2:1, v/v/v) formed the
mobile phase. Quantitative evaluation was performed using special-purpose software. The linearity of the analytical procedure was sustained by the
numerical parameters such as correlation coefficient (0.99520.9980) and standard error of determination (0.030.20). The limits of detection were
found to be within the range of 5.219.34 ng/spot, and the limits of quantification between 10.21 and 18.09 ng/spot. Recovery studies performed
on two levels of concentration gave values between 96.39 and 102.76%. These results show that the regression approach provides rigorous and
realistic detection and quantification limits and as a consequence can be routinely applied to other analytical systems. This method does not require
expensive analytical instruments compared with classical densitometry and provides a reliable quantitative evaluation with minimum of time.
2008 Elsevier B.V. All rights reserved.

Keywords: HP-TLC; Food-dyes; Quantitative evaluation; Image processing; Validation

1. Introduction them involved expensive apparatus. The most common difficul-

Thin-layer chromatography is one of the cheapest, quick-
est and most efficient separation methods for many classes of
chemical compounds. It has some important advantages over
the other chromatographic techniques like low cost of instru-
mentation, evaluation of the whole sample because of the spatial
separation, the ability of making simultaneous separations (even
20 samples) and, of course, shortened time required for anal-
ysis. All these make thin-layer chromatography a convenient
choice for many applications including analytical, bio-medical
and pharmaceutical field.
Scanning densitometry is the classical method for quanti-
tative evaluation of chromatographic plates [13]. It is based
on measuring the absorbance or fluorescense of different zones
on the plate exposed to monochromatic source of light. Many
techniques have been reported in literatures [48], but all of
Corresponding author. Tel.: +40 264 593833; fax: +40 264 590818.
E-mail address: costelsrb@yahoo.co.uk (C. Sarbu).
ties that are encountered here are increased time of scanning
(from 30 min to 2 h) and low resolution. An alternative way
of quantitative evaluation is represented by the charged cou-
pled devices (CCDs) systems [910] that are bidimensional
detectors containing sensors capable of recording an image of
a surface in a matter of seconds. Still, the main limitation is
the unequal illumination of the recorded surface, which leads
to errors [11]. Another solution for obtaining chromatographic
images is to use a flatbed scanner. Owing to the uniform lighting
of surfaces, short time of scanning (approximately 1020 s) and
high optical resolution, it is suitable for recording chromato-
graphic plates. These systems based on digital processing of
chromatographic images are reported in literature as an impor-
tant quantitative method in thin-layer chromatography [1215].
Also, it is possible to save the images in different file for-
mats (tiff, bmp, jpeg) to store them for a long time without
affecting their quality and access them instantly for eventual
revaluation.
In general, the color of food is associated with certain flavors,
thus influencing the perception of the flavor from sweets to wines

0021-9673/$ see front matter 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.chroma.2008.02.077
296 F. Soponar et al. / J. Chromatogr. A 1188 (2008) 295300
Fig. 1. The structures of the studied compounds.
[16]. Based on this, the producers add different substances in
foods in order to simulate the natural color. The variation of the
color of a certain product from one season to another, the effects
of food processing and storage make the addition of artificial
dyes a valuable technique, in this way maintaining the natural
hue or the one preferred by the consumer.
In this paper, three different dyes have been analyzed, which
are encountered in many products on the Romanian market and
not only. Synthetic azo-dyes represent about 6070% of dyes
used in food and textile industry; they are more stable than nat-
ural dyes, are cheap and are highly soluble in water. The only
disadvantage could be the fact that they are insoluble in oils or
organic solvents. As they are highly soluble in water, they do
not remain in the organism (being metabolized in the liver), and
they are eliminated through urine.
A high-performance liquid chromatography method using
a short column with photodiode array detection has been
used in dye analysis (including the studied compounds in this
paper), and it is recommended when the mixture contains many
different colorants [17]. In this case the recoveries ranged
from 76.6 to 115.0%. Another chromatographic method using
high-performance liquid chromatography with UV-diode array
detection (DAD) [18] has been successfully applied for col-
orant analysis including tartrazine and Sunset Yellow in several
commercial food products with better recoveries (98%). UV
molecular absorption spectrometric techniques based on multi-
variate calibration methods have also been reported [19], proving
to be very useful in the determination of tartrazine, Sunset Yel-
low and Ponceau 4R with 94.5105.3% recovery values.
A method that combines high-performance thin-layer chro-
matography with digital processing is proposed for the
determination of three food dyes. The structures are given in
Fig. 1.
2. Experimental
2.1. Reagents
The dyes used in this study (tartrazine, azorubine and Sunset
Yellow) were obtained from Fluka (Buchs, Switzerland). Diethyl
ether and ammonia (concentrated solution) were from Reactivul
(Bucharest, Romania), while isopropanol was obtained from
Silal Trading (Bucharest, Romania). All these solvents had ana-
lytical grade purity. Distilled water was the solvent used for the
dyes.
2.2. Apparatus and software
For quantitative determination a flatbed HP ScanJet 3970 was
used at an optical resolution of 200 dpi. The software for digital
processing (Macherey-Nagel TLC Scanner software) [20] was
developed by Wieczorrek (Macherey-Nagel). The calculation of
limits of detection and quantification was realized with SMAC
(statistical methods in analytical chemistry) [21]. For statistical
data treatment, STATISTICA was used.
Data processing can be made on any computer with a medium
configuration (500 MHz processor, 128 Mb RAM), but for image
processing a configuration with a high video memory and high
frequency processor (3 GHz, 512 Mb RAM) is recommended.
Sample application was done using a 10 L Shimadzu
microsyringe.
2.3. Sample and standard preparation
The analysis of synthetic dyes was performed from five
products, namely five dried concentrated juices, commercially
available in different formats, according to the producer. These
 F. Soponar et al. / J. Chromatogr. A 1188 (2008) 295300
Table 1
Food products which contain the studied dyes
Sample number Commercial name Tartrazine
1 Frutti Orange Present
2 Fruit Orange Present
3 Bolero Cherry Absent
4 Bolero Orange Present
5 Bolero Peach Present
products can be bought from different food-stores. An overview
is available in Table 1.
Three stock solutions of tartrazine, Sunset Yellow and azoru-
bine were prepared by dissolving 2.0019 g, 1.0098 g, and
2.1337 g, respectively in 1000 mL distilled water. These solu-
tions were next used in preparing the standards.
For each dye, six standards were made with a concentra-
tion between 30 and 180 mg L1 for Sunset Yellow, 40.04 and
190.19 mg L1 for tartrazine and 42.68 and 202.73 mg L1 for
azorubine.
To determine the dyes from the products, 23 g of sample
from each product was diluted to a volume of 100 mL and ana-
lyzed without any additional treatment.
2.4. HP-TLC procedure
For separation of dyes Nano-SIL NH2 /UV 254 chro-
matographic plates were used with amino-modified layer
(10 cm 10 cm) from Macherey-Nagel. Precise volumes of
1 L were applied as spots in duplicates at 1 cm above the base
and the edges of the plate using the 10 L microsyringe. After
the plates have been dried in air for 15 min to eliminate any trace
of water, they have been developed by ascending chromatogra-
phy in a developing chamber saturated with vapors of mobile
phase. The developing took place at room temperature.
Mobile phase was a mixture of isopropanol, diethyl ether
and ammonia (solution) at a ratio of 2:2:1 (v/v/v). A volume of
25 mL of mobile phase was enough for each elution.
The developing distance was 5 cm; after the elution the plates
were dried at 60 C and prepared for scanning process.
2.5. Obtaining the chromatographic images and storage
The chromatographic plates were scanned using TrueColor
setting of the HP ScanJet Driver, and the images were saved as
TIFF files without any compression to avoid the loss of image
quality.
3. Results and discussion
3.1. Chromatograms processing
It is common knowledge that the digitalization of images
obtained by flatbed scanners or digital (photo or video) cameras
result in color data being represented by a set of RGB (red,
green, blue) values. These parameters are the color intensities
297
Sunset Yellow Azorubine Producer
Present Absent Kendy (Bulgaria)
Present Absent Alta Vista (Romania)
Absent Present Eurostock (Bulgaria)
Present Absent Eurostock (Bulgaria)
Absent Absent Eurostock (Bulgaria)
of the red (R), green (G) and blue (B) channels [22]. The color
value is calculated by the additive subtraction of RGB intensities
from white.
In Fig. 2a the chromatogram of a mixture of Sunset Yellow
(yelloworange), azorubine (redviolet) and tartrazine (yellow)
can be observed. As said before, the chromatogram was obtained
by additive subtraction of RGB intensities from white. Fig. 2c
shows again the same chromatogram, but this time, adjusted to
its three RGB channels. Therefore, it can be seen that only cer-
tain channels are responsible for the color value of a given point
from the chromatographic plate. For example, in Fig. 2c it can
be noticed that in case of tartrazine only the blue channel con-
tributes to the final chromatographic peak [23], when in the case
of azorubine the blue channel together with the green one can be
taken into consideration. To Sunset Yellow the blue channel has
a major influence with a small contribution from green channel,
as expected, because Sunset Yellow and tartrazine have a simi-
lar color. In all cases the red channel has a low influence on the
final chromatogram, being responsible for eventual impurities
on the plate or for the background noise. The ability to decom-
pose a given peak corresponding to a colored spot into three
components, each one with its contribution of information, can
provide the possibility to choose the component that is more
suitable for analysis. Therefore, the blue channel was selected
for Sunset Yellow because the red and green channels due to
their low content of information were influenced by the back-
ground noise. In this way the final results are less influenced by
errors. This option can be also useful in cases of compounds that
present low resolution. For instance, the calibration for azoru-
bine was done using only the green channel because the blue one
is partly influenced by tartrazine. For tartrazine there is no option
left than to work on the blue channel, also with good results.
Chromatographic analysis through Macherey-Nagel TLC Scan-
ner software implies the detection of the spots by referring to a
clean zone from the developed plate, the selection of appropriate
channel (or combination of channels) corresponding to each dye,
and the integration of the peaks. Further the concentration of the
spot vs. integral is represented, the linear regression equation
is obtained and by interpolation the unknown concentration is
determined.
3.2. Linearity, limits of detection and quantification
The linear domains as well as the calibration curve have
been studied in this paper. In this way 1 L of each stan-
dard was applied in duplicates and analyzed using the method
298 F. Soponar et al. / J. Chromatogr. A 1188 (2008) 295300

Fig. 2. The position of the spots from the solvent front on the chromatographic plate (b) and the chromatograms disposed on all three channels (a) and on separate
channels (c). 1 = Sunset Yellow; 2 = azorubine; 3 = tartrazine.

Table 2
The linear domain, statistic parameters, limits of detection and quantification
Sunset Yellow Tartrazine Azorubine

Linear domain (mg L1 ) 30.03180.18 40.04190.19 42.68202.73

Regression equationa y = 0.602(0.012)x 2.300(1.408) y = 0.291(0.008)x + 6.457(1.060) y = 0.286(0.009)x + 3.176(1.200)
Correlation coefficient, r 0.9980 0.9958 0.9952
Standard error (SE) 0.20 0.03 0.03
F 2503 1194 1028
Limit of detection (ng/spot) 5.21 8.13 9.34
Limit of quantification (ng/spot) 10.21 15.78 18.09

x: sample concentration.
a y: peak integral.

described previously. Linear relation between the applied con-
centrations and the integrals of the peaks is described in
Table 2.
There is a relatively large linear domain, where the correlation
coefficients have values in the range of 0.99520.9980. Limits of
detection and quantification have been calculated using SMAC
software, based on confidence bands generated from calibration
experiments using ordinary least squares method [24,25].
3.3. Precision and accuracy
In order to determine the precision of six identical spots, each
one containing 100.1 ng/spot of Sunset Yellow, 106.7 ng/spot of
azorubine and 130.10 ng/spot of tartrazine, were applied. The
chromatographic plate was developed and scanned as described
before. The data are presented in Table 3, the precision being
estimated in terms of standard deviation.

Table 3
Precision determined by measuring six replicates
Food dye Mean signal (arbitrary units) Standard deviation (SD) Standard error of the mean (SEM)

Sunset Yellow 103.33 1.63 0.67

Azorubine 70 1.67 0.68
Tartrazine 89.5 1.87 0.76
 F. Soponar et al. / J. Chromatogr. A 1188 (2008) 295300 299

Table 4
Recovery assessed with the method of standard additions at two concentration levels
Food dye Initial (ng/spot) Added (ng/spot) Observed (ng/spot) Recovery (%)
Tartrazine 0 120.12 115.78 96.39
21.29 100.10 118.68 97.76
Sunset Yellow 0 150.15 154.30 102.76
22.54 50.05 71.54 98.55
Azorubine 0 128.04 130.70 102.08
34.59 51.45 86.69 101.27

Table 5
Quantitative estimation of the three synthetic azo-dyes
Sample Net weight (g/product) Amount of dye (mg/product) Amount of dye (%)

Tartrazine Sunset Yellow Azorubine Tartrazine Sunset Yellow Azorubine

1 5 19.04 0.92 15.40 0.54 0.38 0.02 0.31 0.01

2 4 11.94 0.82 10.76 0.87 0.3 0.02 0.27 0.02
3 8 36.52 1.49 0.46 0.02
4 8 16.32 1.77 8.22 1.96 0.20 0.02 0.10 0.02
5 8 26.39 1.04 0.33 0.01

The accuracy of the proposed method was determined by
internal standard addition. There have been analyzed solutions
with zero initial concentration, as well as solutions with a known
initial concentration. Precise quantities of pure dyes were added
to 1 g of product sample. For making a solution with zero ini-
tial concentration, the pure dye was diluted with distilled water.
Table 4 contains the data that demonstrate a good recovery of
all the studied dyes.
3.4. Quantitative evaluation
The method being validated through precision, linearity and
accuracy, the next step was the quantitative determination of
the dyes from the commercial product, which is the purpose of
this study. The results of the validation attest the fitness of the
proposed analytical procedure for the identification and quanti-
tative determination of the three dyes in mixture. Five different
commercial products that contained at least one of the three
studied dyes were analyzed in order to assign the values of their
contents. The dyes were simultaneously evaluated, on the same
plate, reducing the time and the amount of the materials required
for the analysis. The obtained results and their confidence inter-
vals are listed in Table 5. As the net weight of the products is
different depending upon the producer, the amount of the dyes
are listed in weight percentages too in order to facilitate the com-
parison of their amounts. The contents of the food dyes are well
correlated with the desirable color of the product that stands on
the natural tint of the fruit (Table 1).
4. Conclusion
In this paper a method for the determination of some syn-
thetic azo-dyes is proposed. Compared with other methods that
obtained similar results but with more expensive and sophisti-
cated apparatus, the instrumentation used in this report is more
simple and easy to handle and acquire. This method combines
high-performance thin-layer chromatography with digital pro-
cessing of images, a domain that is still growing and open to
further improvements. The small cost of the materials, instru-
mentation and a short time of scanning required (compared with
classical densitometry) make this method an optimal choice for
rapid quantitative evaluation of different samples. The validation
of this method demonstrates once again that using appropriate
resources, thin-layer chromatography combined with image pro-
cessing can be a powerful tool in quantitative determination of
dyes in commercial products.
Also, this technique can be applied to other classes of
compounds, colored or colorless, in the last instance making
necessary the use of a UV lamp connected to a digital camera.
References
[1] V. Pollak, J. Schulze-Clewing, J. Planar Chromatogr. 3 (1990) 104.
[2] R. Vijayakannan, M. Karan, S. Dutt, V. Jain, K. Vasisht, Chromatographia
63 (2006) 277.
[3] S.A. Coran, V. Giannellini, M. Bambagiotto-Alberti, J. Chromatogr. A 1045
(2004) 217.
[4] P. Trivedi, K. Pundarikakshudu, Chromatographia 65 (2007) 239.
[5] A. Witek, H. Hopkala, G. Matzsik, Chromatographia 50 (1999) 41.
[6] T. Ohno, E. Mikami, H. Oka, J. Nat. Med. 60 (2006) 141.
[7] N.P. Singh, A.P. Gupta, A.K. Sinha, P.S. Ahuja, J. Chromatogr. A 1077
(2005) 202.
[8] A.S. Stepanov, Pharm. Chem. J. 37 (2003) 54.
[9] J.V. Sweedler, R.B. Bilhorn, P.M. Epperson, G.R. Sims, M.B. Denton, Anal.
Chem. 60 (1988) 282.
[10] J.V. Sweedler, R.B. Bilhorn, P.M. Epperson, G.R. Sims, M.B. Denton, Anal.
Chem. 60 (1988) 372.
[11] I. Vovk, M. Prosek, J. Chromatogr. A 768 (1997) 329.
[12] C.F. Poole, J. Chromatogr. A 856 (1999) 399.
[13] A.V. Gerasimov, Zh. Anal. Khim. 55 (2000) 1292.
[14] A.V. Gerasimov, Zh. Anal. Khim. 59 (2004) 392.
[15] L. Zhang, X. Lin, J. Chromaotgr. A 1109 (2006) 273.
[16] J. Delwiche, Food Qual. Prefer. 15 (2004) 137.
[17] N. Yoshioka, K. Ichihashi, Talanta 74 (2008) 1408.
300 F. Soponar et al. / J. Chromatogr. A 1188 (2008) 295300

[18] S. Pereira Alves, D.M. Brum, E.C. Branco de Andrade, A.D. Pereira Netto,
Food Chem. 107 (2008) 489.
[19] J.J. Berzas Nevado, J. Rodriguez Flores, M.J. Villasenor Llerena, J. Anal.
Chem. 361 (1998) 465.
[20] C. Wieczorrek, J. Planar Chromatogr. 18 (2005) 181.
[21] P.C. Meier, R.E. Zund, Statistical Methods in Analytical Chemistry, Wiley,
New York, 1993.
[22] E.I. Giorgianni, T.E. Madden, Digital Color Management: Encoding Solu-
tions, Addison Wesley, Reading, MA, 1998.
[23] A.V. Gerasomiv, Zh. Anal. Khim. 58 (2002) 241.
[24] C. Sarbu, L. Soran, M. Curtui, Rev. Roum. Chim. 55 (2004) 137.
[25] A. Hubaux, G. Vos, Anal. Chem. 42 (1970) 849.